Gambling involves staking money or other valuables on the outcome of a game or event determined, partially or entirely, by chance. ¹ Today, gambling activities occur in the vast majority of cultures and can be traced back 4,000 years. ² Prevalence studies indicate that almost half of adults worldwide have gambled in the past 12 months. ³ Globally, it is estimated that 8.7% of adults engage in risky gambling, while 1.4% struggle with problematic gambling. ³ The diagnosis of gambling disorder is reserved for the most severe cases and is characterized by impaired control over gambling, an increasing priority given to gambling, and the continuation or escalation of gambling despite negative consequences. ⁴

The global gambling market was valued at approximately $540 billion in 2023, and is expected to grow 6.4% annually, driven by factors such as legalization, urbanization, increased social media use, and a rising population. ⁵ Various technological advancements are anticipated to spur growth in online gambling channels, including gambling apps and online casinos. ^{5, 6} Responsible gambling initiatives aim to promote awareness and prevent harms associated with gambling, but many of these have faced criticism for placing one onus of responsibility on the individuals. ⁷ Wardle et al. ⁸ argue that the growth of the gambling industry is facilitated by strong ties to governments, and emphasize that health related harms of problematic gambling should take precedence over economic interests, commercial profitability, and government revenue generation. Online gambling is expanding rapidly, and in parallel so does the amount of player account data, ⁹ which facilitates the use of various artificial intelligence (AI) technologies. ¹⁰

AI refers to computer systems that automatically perform complex tasks such as perceiving, reasoning, decision making, problem-solving, and learning, which require intelligence when conducted by humans. ^{11– 13} Some AI’s technologies mine data, learn, recognize speech and images, and analyze cognitions and emotions. ¹¹ Machine learning involves computer algorithms that identify patterns in data and enhance their performance automatically over time. ^{14, 15} Among machine learning subsets, deep learning employes a group of machine learning algorithms to execute high-level abstractions in data, resembling human brain functions, using deep architectures or interconnected nodes that constitute multiple non-linear transformations. Deep learning systems automatically learn features at multiple levels of abstraction, enabling them to learn complex functions and map raw sensory input data to the output without human intervention. ¹⁶ Deep learning techniques can be applied in areas such as speech recognition, ¹⁷ object recognition, ¹⁸ and natural language processing. ¹⁹ On the other hand, artificial neural networks, which resemble biological neural networks, perform specific tasks similarly to that of the human brain, using hundreds or thousands of interconnected artificial neurons or nodes that process and transmit information. ^{20, 21} AI has revolutionized automated online service interactions through advancement in algorithms, massive data and affordable computational power and storage. ²² By transforming online consumer behavior into actionable strategies, AI-based digital market businesses can access their customers just-in-time. ^{23– 25} Features like speech recognition and machine learning, enhance mixed reality (MR), immersion, enjoyment and consumers’ perception, positively impacting engagement, purchase intentions and social sharing. ²⁶ Studies also highlight AI’s ability to strengthen relationships between digital markets and online users. ^{27, 28}

In the realm of gambling, AI systems collect and process various forms of consumer data in real time to enhance the understanding of customers (Dos Santos, 2015). For example, models based on neural networks have been developed to predict bet amounts and cumulative winnings/losses. ²⁹ Several studies have used AI to identify problem gamblers. ^{30– 33} In one study, Auer and Griffiths ³⁴ used objective account based data from online casino players and matched it with self-reported gambling problems. Based on data such as wagering, depositing, and gambling frequency, random forest and gradient boost machine algorithms were trained to predict self-reported problem gambling. The two AI algorithms achieved an area under the curve (AUC) value of.73 and.67, respectively, which is considerably better than chance. ³⁴ Other studies using machine learning models have exclusively used account based data, showing that early behaviors (within the first 7 days) can predict later high risk gambling. ³⁵ Machine learning models, exclusively based on survey data have shown that reports of gambling problems can be predicted by various demographic and self-reported gambling-related behaviors. ³⁶ Suicidal ideation and suicide attempts among gamblers have also been modelled with machine learning using similar data. ³⁷ In addition, AI has also been used to identify personality traits that distinguish between gambling disorder patients and healthy controls. ³⁸ One study used mixed logistic regression machine learning based on data from four bet types and eight seasons of English Premier League soccer, finding that losses were positively correlated with observable betting odds. The authors suggest incorporating differences in product risk into consumer education and responsible gambling approaches. ³⁹ Another application of AI within the gambling sphere concerns predicting outcomes in skill-based games (e.g., poker). Studies have shown that AI can outperform top poker players in some cases. ^{40, 41} Research has also been conducted to assess and validate the skills of poker players against AI-simulated games. ⁴² A qualitative interview study with gambling industry stakeholders suggests that while AI holds promise for consumer protection, there is also an inherent risk of increased player exploitation (e.g., creating more addictive games and more persuasive marketing tactics). ⁴³

Based on this backdrop, the aim of this scoping review is to map studies where AI-based tools have been used within the gambling field, areas of use as well as current trends and findings.

AI is used increasingly within the gambling industry. Based on both subjective data as well as player account data AI holds promise as a tool for predicting future gambling behavior, identifying those with problems as well as validating specific skills involved in gambling. Within the gambling context AI may thus be beneficiary for customers, however the potential for misuse in terms of customer exploitation cannot be dismissed. We regard the scoping review approach as the best suited method to gain an overview and to make a synthesis regarding how AI currently is being used within the gambling industry as well as pointing out future and coming trends. The review will inform various stakeholders (e.g., regulatory agencies, the industry, clinicians as well as various user organization) about the use of AI within the gambling field. A limitation of the planned scoping review is that it will be restricted to publications in English. In addition, unpublished paper/report made by the industry will not be included.